What Do You Look at First When Making a Football Prediction?

[connorhayes47]

When analysing a football match, which factors do you normally give the most weight?

I usually start with recent form and home/away performance, but injuries, suspensions and tactical matchups can sometimes matter more than the raw results.

A few things I normally check:

- Recent form
- Home and away records
- Injuries and suspensions
- Head-to-head results
- Expected starting lineups
- Motivation
- Schedule congestion

Do you rely mainly on statistics, or do you also put a lot of weight on actually watching the teams?

Interested to hear how other people approach football predictions.

 

[Klaus]

When analysing a football match, which factors do you normally give the most weight?

I usually start with recent form and home/away performance, but injuries, suspensions and tactical matchups can sometimes matter more than the raw results.

A few things I normally check:

- Recent form
- Home and away records
- Injuries and suspensions
- Head-to-head results
- Expected starting lineups
- Motivation
- Schedule congestion

Do you rely mainly on statistics, or do you also put a lot of weight on actually watching the teams?

Interested to hear how other people approach football predictions.

The list you’ve provided is a reasonable starting point. Most of what’s there is standard. The question is how you weight and combine these factors rather than whether you include them.

The most common mistake at the recreational level: treating all factors as roughly equal and forming a general impression rather than a structured output.

My approach for Bundesliga matches is model-based. Expected goals data from recent matches, weighted for opponent quality and home or away context. The model produces a fair probability. I compare that probability to the available market price. If the gap exceeds my threshold, I bet. If not, I pass.

On your specific factors.

Recent form: useful but easily misread. A team winning three consecutive matches may have a lower underlying xG performance than a team losing two. The scoreline flatters one and punishes the other. Raw results without quality context are often noise.

Head-to-head: I weight this less than most. The squads, managers and contexts from previous meetings are often sufficiently different to make historical head-to-head data unreliable. There are exceptions for specific stylistic matchups that persist across squad changes.

Motivation: significant and systematically underweighted by standard models. A mid-table team with nothing to play for facing a top-half side in a congested fixture list is a different match from what the quality metrics suggest. I incorporate this as a manual adjustment.

On watching versus statistics.

Both are necessary. Statistics without watching misses tactical context. Watching without statistics produces confirmation bias. The model tells you what the numbers suggest. Watching tells you whether the numbers are capturing what’s actually happening.

The combination matters. Neither alone is sufficient.

 

[derbyScala]

Form is the first thing most people open, and it is also the one that lies most often. A 2-0 that was 0.4 xG and a 1-1 that was 2.8 look the same on the results page. Home and away splits help until the sample is six games. Injuries and the expected XI are the ones that actually move a price between Thursday and Saturday. H2H is the factor that gets more weight than it deserves once both squads have changed. Motivation and congestion are the ones models miss and eyes catch. Watching the last two games after the numbers are done is the useful order. Watching first and then finding the numbers that agree is the other one.

 

[oddsfactory]

Agree with Klaus and derbyScala that raw form is the least reliable item on the list. A 2-0 at 0.4 xG and a 1-1 at 2.8 are very different teams. Most of the other factors (injuries, motivation, congestion) really feed into the expected XI, and that's the one input that can change after a pick is published. Klaus, on motivation as a manual adjustment: do you apply a fixed shift to the probability or scale it per match? And have you checked whether those adjustments added or lost value compared with the unadjusted model? (Disclosure: I run an AI football picks site that uses Poisson plus an LLM for news context, so I'm asking for practical reasons.)

 

[patchworkfox]

When analysing a football match, which factors do you normally give the most weight?

I usually start with recent form and home/away performance, but injuries, suspensions and tactical matchups can sometimes matter more than the raw results.

A few things I normally check:

- Recent form
- Home and away records
- Injuries and suspensions
- Head-to-head results
- Expected starting lineups
- Motivation
- Schedule congestion

Do you rely mainly on statistics, or do you also put a lot of weight on actually watching the teams?

Interested to hear how other people approach football predictions.

Whatever factors you use, the step that makes them useful is turning your read into a probability and comparing it with the market. To do that fairly you have to strip the bookmaker margin out first. Example with 1X2 odds of 2.10 / 3.40 / 3.60: the implied probabilities are 47.6%, 29.4% and 27.8%, which add up to 104.8%. Divide each by 1.048 and the fair home price is about 45.4%, or 2.20 in decimal odds. If your own assessment of form, lineups and motivation puts the home side clearly above 45%, there is something to look at; if it is about the same, the factors just confirm what the price already says.

 

[patchworkfox]

When analysing a football match, which factors do you normally give the most weight?

I usually start with recent form and home/away performance, but injuries, suspensions and tactical matchups can sometimes matter more than the raw results.

A few things I normally check:

- Recent form
- Home and away records
- Injuries and suspensions
- Head-to-head results
- Expected starting lineups
- Motivation
- Schedule congestion

Do you rely mainly on statistics, or do you also put a lot of weight on actually watching the teams?

Interested to hear how other people approach football predictions.

Whatever factors you use, the step that makes them useful is turning your read into a probability and comparing it with the market. To do that fairly you have to strip the bookmaker margin out first. Example with 1X2 odds of 2.10 / 3.40 / 3.60: the implied probabilities are 47.6%, 29.4% and 27.8%, which add up to 104.8%. Divide each by 1.048 and the fair home price is about 45.4%, or 2.20 in decimal odds. If your own assessment of form, lineups and motivation puts the home side clearly above 45%, there is something to look at; if it is about the same, the factors just confirm what the price already says.

 

[TipsAudit]

I start at the other end: the sharpest price I can find, with the margin taken out, because it already contains most of that list. Then I ask what I know that it doesn't.

We tested this on our own data: about 50 experiments with Elo, form, xG, lineups and missing players, run walk-forward so nothing leaked. On backtest, none of them beat the closing price of a sharp book. Missing players were the clearest case. The effect is real and you can measure it, but in the leagues where we have historical prices it was already in the odds.

The market's own chances also hold up when you check them. In the record we keep of its favourites since 18 October, 345 of 480 won (72%), at an average chance of 73%.

So my order is price first, then the expected XI if it isn't out yet, because that is the one item on your list that can still move the number. Form and head-to-head come last, since they are the most public. Motivation I would only use if you have tracked whether your adjustments actually helped, which is what oddsfactory asked Klaus above.

Disclosure: I run TipsAudit, which measures this kind of thing, so weigh that accordingly.